Get the FREE Ultimate OpenClaw Setup Guide →
npx machina-cli add skill Orchestra-Research/AI-Research-SKILLs/pytorch-lightning --openclaw
Files (1)
SKILL.md
8.9 KB

PyTorch Lightning - High-Level Training Framework

Quick start

PyTorch Lightning organizes PyTorch code to eliminate boilerplate while maintaining flexibility.

Installation:

pip install lightning

Convert PyTorch to Lightning (3 steps):

import lightning as L
import torch
from torch import nn
from torch.utils.data import DataLoader, Dataset

# Step 1: Define LightningModule (organize your PyTorch code)
class LitModel(L.LightningModule):
    def __init__(self, hidden_size=128):
        super().__init__()
        self.model = nn.Sequential(
            nn.Linear(28 * 28, hidden_size),
            nn.ReLU(),
            nn.Linear(hidden_size, 10)
        )

    def training_step(self, batch, batch_idx):
        x, y = batch
        y_hat = self.model(x)
        loss = nn.functional.cross_entropy(y_hat, y)
        self.log('train_loss', loss)  # Auto-logged to TensorBoard
        return loss

    def configure_optimizers(self):
        return torch.optim.Adam(self.parameters(), lr=1e-3)

# Step 2: Create data
train_loader = DataLoader(train_dataset, batch_size=32)

# Step 3: Train with Trainer (handles everything else!)
trainer = L.Trainer(max_epochs=10, accelerator='gpu', devices=2)
model = LitModel()
trainer.fit(model, train_loader)

That's it! Trainer handles:

  • GPU/TPU/CPU switching
  • Distributed training (DDP, FSDP, DeepSpeed)
  • Mixed precision (FP16, BF16)
  • Gradient accumulation
  • Checkpointing
  • Logging
  • Progress bars

Common workflows

Workflow 1: From PyTorch to Lightning

Original PyTorch code:

model = MyModel()
optimizer = torch.optim.Adam(model.parameters())
model.to('cuda')

for epoch in range(max_epochs):
    for batch in train_loader:
        batch = batch.to('cuda')
        optimizer.zero_grad()
        loss = model(batch)
        loss.backward()
        optimizer.step()

Lightning version:

class LitModel(L.LightningModule):
    def __init__(self):
        super().__init__()
        self.model = MyModel()

    def training_step(self, batch, batch_idx):
        loss = self.model(batch)  # No .to('cuda') needed!
        return loss

    def configure_optimizers(self):
        return torch.optim.Adam(self.parameters())

# Train
trainer = L.Trainer(max_epochs=10, accelerator='gpu')
trainer.fit(LitModel(), train_loader)

Benefits: 40+ lines → 15 lines, no device management, automatic distributed

Workflow 2: Validation and testing

class LitModel(L.LightningModule):
    def __init__(self):
        super().__init__()
        self.model = MyModel()

    def training_step(self, batch, batch_idx):
        x, y = batch
        y_hat = self.model(x)
        loss = nn.functional.cross_entropy(y_hat, y)
        self.log('train_loss', loss)
        return loss

    def validation_step(self, batch, batch_idx):
        x, y = batch
        y_hat = self.model(x)
        val_loss = nn.functional.cross_entropy(y_hat, y)
        acc = (y_hat.argmax(dim=1) == y).float().mean()
        self.log('val_loss', val_loss)
        self.log('val_acc', acc)

    def test_step(self, batch, batch_idx):
        x, y = batch
        y_hat = self.model(x)
        test_loss = nn.functional.cross_entropy(y_hat, y)
        self.log('test_loss', test_loss)

    def configure_optimizers(self):
        return torch.optim.Adam(self.parameters(), lr=1e-3)

# Train with validation
trainer = L.Trainer(max_epochs=10)
trainer.fit(model, train_loader, val_loader)

# Test
trainer.test(model, test_loader)

Automatic features:

  • Validation runs every epoch by default
  • Metrics logged to TensorBoard
  • Best model checkpointing based on val_loss

Workflow 3: Distributed training (DDP)

# Same code as single GPU!
model = LitModel()

# 8 GPUs with DDP (automatic!)
trainer = L.Trainer(
    accelerator='gpu',
    devices=8,
    strategy='ddp'  # Or 'fsdp', 'deepspeed'
)

trainer.fit(model, train_loader)

Launch:

# Single command, Lightning handles the rest
python train.py

No changes needed:

  • Automatic data distribution
  • Gradient synchronization
  • Multi-node support (just set num_nodes=2)

Workflow 4: Callbacks for monitoring

from lightning.pytorch.callbacks import ModelCheckpoint, EarlyStopping, LearningRateMonitor

# Create callbacks
checkpoint = ModelCheckpoint(
    monitor='val_loss',
    mode='min',
    save_top_k=3,
    filename='model-{epoch:02d}-{val_loss:.2f}'
)

early_stop = EarlyStopping(
    monitor='val_loss',
    patience=5,
    mode='min'
)

lr_monitor = LearningRateMonitor(logging_interval='epoch')

# Add to Trainer
trainer = L.Trainer(
    max_epochs=100,
    callbacks=[checkpoint, early_stop, lr_monitor]
)

trainer.fit(model, train_loader, val_loader)

Result:

  • Auto-saves best 3 models
  • Stops early if no improvement for 5 epochs
  • Logs learning rate to TensorBoard

Workflow 5: Learning rate scheduling

class LitModel(L.LightningModule):
    # ... (training_step, etc.)

    def configure_optimizers(self):
        optimizer = torch.optim.Adam(self.parameters(), lr=1e-3)

        # Cosine annealing
        scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(
            optimizer,
            T_max=100,
            eta_min=1e-5
        )

        return {
            'optimizer': optimizer,
            'lr_scheduler': {
                'scheduler': scheduler,
                'interval': 'epoch',  # Update per epoch
                'frequency': 1
            }
        }

# Learning rate auto-logged!
trainer = L.Trainer(max_epochs=100)
trainer.fit(model, train_loader)

When to use vs alternatives

Use PyTorch Lightning when:

  • Want clean, organized code
  • Need production-ready training loops
  • Switching between single GPU, multi-GPU, TPU
  • Want built-in callbacks and logging
  • Team collaboration (standardized structure)

Key advantages:

  • Organized: Separates research code from engineering
  • Automatic: DDP, FSDP, DeepSpeed with 1 line
  • Callbacks: Modular training extensions
  • Reproducible: Less boilerplate = fewer bugs
  • Tested: 1M+ downloads/month, battle-tested

Use alternatives instead:

  • Accelerate: Minimal changes to existing code, more flexibility
  • Ray Train: Multi-node orchestration, hyperparameter tuning
  • Raw PyTorch: Maximum control, learning purposes
  • Keras: TensorFlow ecosystem

Common issues

Issue: Loss not decreasing

Check data and model setup:

# Add to training_step
def training_step(self, batch, batch_idx):
    if batch_idx == 0:
        print(f"Batch shape: {batch[0].shape}")
        print(f"Labels: {batch[1]}")
    loss = ...
    return loss

Issue: Out of memory

Reduce batch size or use gradient accumulation:

trainer = L.Trainer(
    accumulate_grad_batches=4,  # Effective batch = batch_size × 4
    precision='bf16'  # Or 'fp16', reduces memory 50%
)

Issue: Validation not running

Ensure you pass val_loader:

# WRONG
trainer.fit(model, train_loader)

# CORRECT
trainer.fit(model, train_loader, val_loader)

Issue: DDP spawns multiple processes unexpectedly

Lightning auto-detects GPUs. Explicitly set devices:

# Test on CPU first
trainer = L.Trainer(accelerator='cpu', devices=1)

# Then GPU
trainer = L.Trainer(accelerator='gpu', devices=1)

Advanced topics

Callbacks: See references/callbacks.md for EarlyStopping, ModelCheckpoint, custom callbacks, and callback hooks.

Distributed strategies: See references/distributed.md for DDP, FSDP, DeepSpeed ZeRO integration, multi-node setup.

Hyperparameter tuning: See references/hyperparameter-tuning.md for integration with Optuna, Ray Tune, and WandB sweeps.

Hardware requirements

  • CPU: Works (good for debugging)
  • Single GPU: Works
  • Multi-GPU: DDP (default), FSDP, or DeepSpeed
  • Multi-node: DDP, FSDP, DeepSpeed
  • TPU: Supported (8 cores)
  • Apple MPS: Supported

Precision options:

  • FP32 (default)
  • FP16 (V100, older GPUs)
  • BF16 (A100/H100, recommended)
  • FP8 (H100)

Resources

Source

git clone https://github.com/Orchestra-Research/AI-Research-SKILLs/blob/main/08-distributed-training/pytorch-lightning/SKILL.mdView on GitHub

Overview

PyTorch Lightning is a high-level framework that organizes PyTorch code around a Trainer and LightningModule, eliminating boilerplate while preserving flexibility. It brings built-in best practices, including automatic distributed training (DDP/FSDP/DeepSpeed), mixed precision, checkpointing, logging, and progress bars, enabling scalable training from laptop to supercomputer with the same code.

How This Skill Works

Developers implement a LightningModule defining training_step, validation_step, test_step, and configure_optimizers. The Trainer then handles device placement, distributed strategies, precision, gradient accumulation, checkpointing, and logging, so the same code runs on CPU, single/multi-GPU, or TPU without manual boilerplate.

When to Use It

  • You want clean, boilerplate-free training loops with built-in best practices.
  • You need to scale training across GPUs or nodes with minimal code changes (DDP/FSDP/DeepSpeed).
  • You want automatic mixed precision (FP16/BF16) and device management.
  • You prefer integrated logging, checkpointing, and progress bars without writing extra boilerplate.
  • You're migrating PyTorch code and want a structured, maintainable workflow.

Quick Start

  1. Step 1: Define a LightningModule with training_step (and optionally validation_step) and configure_optimizers.
  2. Step 2: Create your DataLoader (train_loader) and any validation/test loaders as needed.
  3. Step 3: Train with Trainer, e.g. trainer = L.Trainer(max_epochs=10, accelerator='gpu', devices=2); trainer.fit(LitModel(), train_loader).

Best Practices

  • Organize model code in a LightningModule with training_step, validation_step, and configure_optimizers.
  • Use self.log(...) for metrics to ensure proper logging across devices and loggers.
  • Let Trainer manage devices, distributed training, and precision rather than manual .to calls.
  • Leverage built-in features like checkpointing and callbacks for early stopping and model selection.
  • Start with CPU/local tests, then scale to GPUs with small batches and short epochs before full-scale runs.

Example Use Cases

  • Convert PyTorch to Lightning by wrapping your model in a LitModel with training_step and configure_optimizers, then train with trainer.fit.
  • Train on multiple GPUs: trainer = L.Trainer(max_epochs=10, accelerator='gpu', devices=2); trainer.fit(LitModel(), train_loader).
  • Add validation and testing: implement validation_step and test_step, log val_loss and val_acc during training.
  • Automatic device handling: no explicit .to('cuda') calls needed; Lightning maps batches to the correct device automatically.
  • Enable FP16/mixed precision to save memory and speed up training using Trainer's precision features.

Frequently Asked Questions

Add this skill to your agents

Related Skills

huggingface-accelerate

Orchestra-Research/AI-Research-SKILLs

Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.

deepspeed

Orchestra-Research/AI-Research-SKILLs

Expert guidance for distributed training with DeepSpeed - ZeRO optimization stages, pipeline parallelism, FP16/BF16/FP8, 1-bit Adam, sparse attention

moe-training

Orchestra-Research/AI-Research-SKILLs

Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scaling model capacity without proportional compute increase. Covers MoE architectures, routing mechanisms, load balancing, expert parallelism, and inference optimization.

pytorch-fsdp2

Orchestra-Research/AI-Research-SKILLs

Adds PyTorch FSDP2 (fully_shard) to training scripts with correct init, sharding, mixed precision/offload config, and distributed checkpointing. Use when models exceed single-GPU memory or when you need DTensor-based sharding with DeviceMesh.

ray-train

Orchestra-Research/AI-Research-SKILLs

Distributed training orchestration across clusters. Scales PyTorch/TensorFlow/HuggingFace from laptop to 1000s of nodes. Built-in hyperparameter tuning with Ray Tune, fault tolerance, elastic scaling. Use when training massive models across multiple machines or running distributed hyperparameter sweeps.

ray-data

Orchestra-Research/AI-Research-SKILLs

Scalable data processing for ML workloads. Streaming execution across CPU/GPU, supports Parquet/CSV/JSON/images. Integrates with Ray Train, PyTorch, TensorFlow. Scales from single machine to 100s of nodes. Use for batch inference, data preprocessing, multi-modal data loading, or distributed ETL pipelines.

Sponsor this space

Reach thousands of developers